Patients' recollections of their past symptoms, illnesses, and episodes of care are often inconsistent from one inquiry to the next. Patients frequently fail to recall (and therefore underreport) the incidence of previous symptoms and events; tend to combine separate, similar occurrences into a single, generic memory; and falsely recall medical events and symptoms that did not in fact occur. This unreliability of recall is affected by personality characteristics and by the patient's current state at the time of recall. Thus, current anxiety or depression and pain or bodily distress foster the recall of symptoms and events that are not recalled when the patient is more comfortable. Finally, current beliefs about one's health and the nature and causes of one's illness also affect the recall of past symptoms and illness. Physicians can maximize the reliability of the clinical history by (1) noting and taking into account the patient's current physical and emotional state; (2) first establishing historical "anchor points" or memorable milestones; (3) decomposing generic memories by finding features that distinguish them from each other; and (4) recalling the clinical history in retrograde fashion, beginning with the most recent event and working backward.A 46-year-old man, when seen for his biannual routine checkup, denied any history of abdominal pain on review of systems. Four weeks later, he returned for an urgent, unscheduled visit to the same physician because of increasingly severe spasmodic abdominal pain of 3 hours' duration. At that time, he gave a history of 3 brief, self-limited episodes of similar but much milder pain occurring approximately 3, 2, and 1 year previously. An abdominal computed tomographic scan at the time of his urgent visit revealed multiple renal stones of various sizes.Patients and healthy individuals alike are surprisingly inconsistent in recalling past symptoms, illnesses, and medical care. When asked the same questions repeatedly over time, they forget the clinical history that they provided earlier and recall information that they had not previously provided. The implications of this unreliability have been considered for epidemiologic, cognitive psychologic, and psychiatric diagnosis, but their ramifications for medical practice have not been widely appreciated.The unreliability of the clinical history has been studied in various ways: (1) patients' recall of symptoms and illnesses has been compared with their medical records for the same period; (2) individuals have used diaries and logs to record symptoms and illnesses in real time and then subsequently have been asked to recall these symptoms; (3) individuals have been asked the same historical questions in an interview or via self-report questionnaire repeatedly over time and their responses have been compared; and (4) psychological experiments have been conducted in which the conditions of learning and recall were carefully controlled or manipulated.The unreliability of memory for clinical events has been amply demonstrated. In general, the lifetime recall of somatic symptoms considerably underestimates their true incidence. For example, when a large sample of primary care patients underwent the same structured interview on 2 occasions 1 year apart, 43% of the somatic symptoms reported initially were not recalled 12 months later.In addition, many more symptoms were reported for the first time at the follow-up interview than would be expected to have occurred during the intervening year, that is, more than half of the lifetime somatic symptoms reported at follow-up had been denied a year earlier.In another study,health maintenance organization patients' recall of all outpatient visits made within a 12-month period were compared with their medical records. Patients recalled 41% of the visits noted in their records, and there was also a 28% incidence of false recall in which the patient recalled a visit for which there was no record.Epidemiologic surveys have yielded comparable findings. Harlow and Linetreviewed several studies in which self-reports of chronic illness, medical care utilization, and medication use were compared with medical record information for periods up to 1 year. Respondents reported 30% to 53% of the diagnoses recorded by their physicians, and physician-recorded diagnoses constituted 36% to 70% of the diagnoses reported by the respondents.Respondents tended to be fairly accurate in recalling hospitalizations but had poor recall of radiologic procedures and ambulatory visits.This failure to recall past morbidity in cross-sectional studies may lead to significant underestimation of lifetime prevalence of disorders and incorrect estimations of temporal trends.In this article, I review the unreliability in somatic symptom recall, describe some factors known to affect it, and suggest some implications for clinical practice.SOURCES OF INSTABILITYWhen the same individual is reinterviewed over time and provides inconsistent or contradictory information, there are at least 3 major sources: patient variance, in which the individual answers the same question differently on different occasions; information variance, in which the interviewer poses the question differently on different occasions (eg, "You didn't have any pain, did you?" vs "Did you have any pain?" or "Did they do an electrocardiogram?" vs "Did they do any tests?"); and interviewer variance, in which the interviewer apprises or interprets the same response differently on different occasions (as is likely to happen with imprecise or ambiguous responses such as "not much" or "only a few times").Patient variance is the most carefully studied source of contradictory information and is the main focus of this article. This does not imply that the physician is not a major source of instability in history taking. Physicians commonly, for example, ask leading questions or force patients to choose from among a limited number of responses to their questions, thereby distorting the patient's story. Patient variance takes several forms. First, as noted, patients tend to forget and therefore underreport the incidence of previous symptoms, illnesses, and medical care interactions. Memory for dates is particularly unreliable. In an event-dating study, some patients misdated an event that had occurred within the past 4 months by as much as 3 months.In another study,patients reported that 20% of their attempts to date autobiographical events were pure guesswork. Forgetting may be a particular problem when the information in question is embarrassing or socially disapproved of, has a pejorative connotation, or is stigmatized, such as psychiatric treatment, alcohol abuse, or sexually transmitted diseases. Second, patients have a tendency to combine multiple, separate incidents or episodes into a single one, a process referred to as "recomposition." That is, incidents with similar characteristics that recur repeatedly tend to be merged into a single, generic memory for the group as a whole.Third, there is the phenomenon of "forward telescoping" in which remote events tend to be displaced forward in time and are remembered as occurring more recently than they actually did.This leads to overreporting of the frequency of events in more recent periods. Finally, patients falsely remember ("fabricate") events and incidents that did not in fact occur. Although fabrication is less frequent than forgetting, it nonetheless occurs often.COMMON CAUSES OF PATIENT VARIANCEPersonality CharacteristicsA personality trait called "negative affectivity" is associated with the overreporting of past illnesses, symptoms, and medical care use. Negative affectivity is a dispositional tendency to experience psychological distress in general, to emphasize the negative aspects of one's experience, and to be critical of oneself and others. Negative affectivity encompasses low morale, poor self-esteem, hypersensitivity, and unpleasant emotions such as guilt, anger, fear, and depression. Individuals with high negative affectivity tend to recall more somatic symptoms and more intense symptoms in retrospect than they recorded in contemporaneous diaries of their symptoms as they occurred in real time.State-Dependent RecallRecall of previous symptoms and illness episodes is affected by the patient's emotional state at the time of recall. In general, people alter their reports of the past to make them consistent with their current state.Thus, patients are more likely to recall somatic symptoms at a follow-up interview that they did not report 1 year earlier if they have an anxiety or depressive disorder at the time of the follow-up interview.Anxiety and depression facilitate the recall of unpleasant past events and negative experiences in general and of illnesses in particular.For example, among individuals with the flu, those who underwent an unpleasant mood induction recalled more symptoms than equally ill individuals who underwent a positive mood induction.Mood affects the perception of symptoms and the appraisal of one's health status.A negative mood state makes illness-related memories more accessible and induces a poorer assessment of one's overall state of health.The current state of one's physical health also affects recall of previous health events and symptoms, although the empirical evidence for this is limited primarily to studies of pain. This memory of past pain is a function of the intensity of currentpain at the time of recall.The clinical vignette at the beginning of this article illustrates how current pain can prompt the recall of similar pain that does not come to mind when the individual is asymptomatic. This effect can be particularly salient when attempting to titrate the dose of an analgesic over time in patients with chronic pain. Extensive literature on patients with chronic pain discloses systematic distortions of memory such that current pain causes one to recall past pain as more severe than it was when it was being experienced.For example, patients with chronic headache noted past pain as more intense when they were currently in pain than when queried during a period of less intense pain, at which time they remembered past pain as milder than they did when they experienced it.This overestimation of past pain occurs more in patients with chronic pain who have more emotional distress and more conflict and discord in their households.This phenomenon has also been explored experimentally. Patients were asked to keep daily diaries of pain intensity for a week. At the end of that time, they received physical therapy to decrease their pain and were asked either before or after the therapy to recall their pain over the preceding week. Those queried immediately after pain relief recalled less pain in the preceding week and recalled taking fewer pain medications during that period (although the groups did not in fact differ on these measures).In another study, when asked to recall a series of words, patients with chronic pain recalled more pain-related words, but not more neutral words, than did control subjects.Illness BeliefsRecall of events may also vary because of changes in the patient's beliefs about, understanding of, or opinions of those events. More specifically, beliefs about one's state of health and about the presumed causes of one's symptoms profoundly affect recall; in other words, the patient's memory of past illnesses and symptoms is affected by beliefs about their cause, course, and treatment.The belief that one is sick causes an underestimation of symptoms in the period before the patient believes he or she became ill and an overestimation of symptoms after that time. Patients with whiplash have thus been found to underestimate their preinjury history of neck symptoms compared with the prevalence of neck symptoms in uninjured controls.Women who believe menstruation is a negative experience recall past menstrual periods as more symptomatic than they reported them to be at the time they were experiencing them compared with women with less negative views of menstrual distress.The 2 groups do not differ, however, in their recall of intermenstrual symptoms.Likewise, informing healthy volunteers in an experiment that they have just tested positive for a disease causes them to recall symptoms that they had previously been told characterized that disease and to recall more behaviors that were described as risk factors for this hypothetical disease.Information-Specific CharacteristicsThe characteristics of the information being recalled also affect the reliability and stability of recall. The more remote in time the symptom or episode, the more unreliable and unstable is its recall; the accuracy of reporting hospital stays, physician visits, and medical conditions and injuries declines as the length of the retention interval increases.Frequent events that are similar in nature, as noted earlier, tend to be "recomposed" into a single event. Respondents are less reliable when recalling information that is perceived as personally threatening or socially undesirable, for example, the number of sexual partners, recreational drug use, or an abortion.Finally, the seriousness of the event, and its singularity or novelty, makes it more likely to be recalled consistently and reliably.CLINICAL IMPLICATIONSSymptoms function as guideposts to diagnosis and benchmarks of treatment, and the unreliability of recall is salient in the diagnostic and treatment processes. For example, when evaluating the palliative effect of a new therapeutic regimen, or attempting to obtain a clear, longitudinal picture of an episodic illness, it is important to remember that the patient's recall of symptoms since the previous visit may actually depend on how severe the symptoms are at the time of the inquiry, how anxious or depressed he or she currently is, and whether he or she has acquired more ominous or worrying beliefs about his or her illness since the previous visit.There are several implications for clinicians. First, information variance should be minimized by ensuring that the history is taken in the same manner each time, questioning the patient in precisely the same terms on every occasion. (There are of course many other instances in which the physician's objective is not to minimize information variance but rather to obtain the patient's story in as unstructured and unconstrained form as possible, ie, in "his or her own words.") Second, clinicians should strive to reduce interviewer variance by interpreting patient responses consistently and by clarifying and qualifying ambiguous or imprecise answers such as "a lot," "all the time," or "several," since such vague language is subject to variable interpretation. Third, this review underscores the importance of noting the patient's current state at the time the history is taken: Is the patient appreciably more or less depressed or anxious than when previously questioned? How physically ill does he or she feel? How much pain or discomfort is he or she currently in? Has the patient's understanding of his or her condition or attribution for his or her symptoms changed? All of these, as we have seen, are likely to affect recall in predictable ways.Several techniques may be helpful. First, recall can be made more reliable by establishing "anchor points."Anchor points are important, memorable, personal, or cultural events (such as a graduation, beginning a new job, or a major news event) that occurred during the period in question. They are helpful because patients remember health events more easily the closer they fall to landmark events and to the boundaries between periods.The respondent can then use these anchors as orientation points and landmarks around which to situate symptoms and clinical events. Another aid to reliability is assisting patients to decompose generic memories (eg, a series of hospitalizations for the same illness or recurrent episodes of the same illness) by finding features that distinguish them from one another.This creates additional cues to retrieval by breaking the task down into smaller units. For instance, past physician visits can be remembered more reliably if they are first broken down into various specialties and subspecialities.Finally, asking patients to recall a sequence of events in the reverse order (beginning with the most recent and working backwards) also aids reliability.In conclusion, the recall of past symptoms, illnesses, and medical care can be unreliable and inconsistent. Given that the clinical history is so crucial to diagnosis and treatment, this phenomenon deserves further scrutiny and mandates that physicians develop techniques to improve patient recall of past medical events.GESimonOGurejeStability of somatization disorder and somatization symptoms among primary care patients.Arch Gen Psychiatry.1999;56:90-95.BMeansANigamMZarrowEFLoftusMSDonaldsonAutobiographical memory for health-related events.Vital Health Stat.1989;6:1-37.SDHarlowMSLinetAgreement between questionnaire data and medical records: the evidence for accuracy of recall.Am J Epidemiol.1989;129:233-248.GESimonMVon KorffRecall of psychiatric history in cross-sectional surveys: implications for epidemiologic research.Epidemiol Rev.1995;17:221-227.DRThompsonMemory for unique personal events: the Roommate Study.Mem Cognit.1982;10:324-332.NMBradburnTemporal representation and event dating.In: Stone AA, Turkkan JS, Bachrach CA, Jobe JB, Kurtzman HS, Cain VS, eds. The Science of Self-Report: Implications for Research and Practice.Mahwah, NJ: Lawrence Erlbaum Associates; 2000:49-61.CThompsonJSkowronskiDLeeReconstructing the date of a personal event.In: Gruenberg M, Morris P, Sykes R, eds. Practical Aspects of Memory: Current Research and Issues. New York, NY: John Wiley & Sons Inc; 1988:241-246. Memory in Everyday Life;vol 1.RJLarsenNeuroticism and selective encoding and recall of symptoms: evidence from a combined concurrent-retrospective study.J Pers Soc Psychol.1992;62:480-488.SCohenJMGwaltney JrWJDoyleDPSkonerPFiremanJTNewsomState and trait negative affect as predictors of objective and subjective symptoms of respiratory viral infections.J Pers Soc Psychol.1995;68:159-169.CSAneshenselALEstradaMJHansellVAClarkSocial psychological aspects of reporting behavior: lifetime depressive episode reports.J Health Soc Behav.1987;28:232-246.GESimonMVon KorffSomatization and psychiatric disorder in the NIMH epidemiologic catchment area study.Am J Psychiatry.1991;148:1494-1500.KKroenkeRLSpitzerJBWWilliamsPhysical symptoms in primary care: predictors of psychiatric disorders and functional impairment.Arch Fam Med.1994;3:774-779.TAAhlesEBBlanchardHLeventhalCognitive control of pain: attention to the sensory aspects of the cold pressor stimulus.Cognit Ther Res.1983;7:159-178.RTCroyleMBUretskyEffects of mood on self-appraisal of health status.Health Psychol.1987;6:239-253.GHBowerMood and memory.Am Psychol.1981;36:129-148.MClarkAIsenToward understanding the relationship between feeling states and social behavior.In: Mastor FA, Sen A, eds. Cognitive Social Psychology.Amsterdam, the Netherlands: Elsevier Science Publishers; 1982:73-108.LCohenLTowbesRFloccoEffects of induced mood on self-reported life events and perceived and received social support.J Pers Soc Psychol.1988;55:669-674.PSaloveyDBirnbaumInfluence of mood on health-relevant cognitions.J Pers Soc Psychol.1989;57:539-551.JWrightSMorleyAutobiographical memory and chronic pain.Br J Clin Psychol.1995;34:255-265.EEichJLReevesBJaegerSBGraff-RadfordMemory for pain: relation between past and present pain intensity.Pain.1985;23:375-379.RNJamisonTSbroccoWCVParrisThe influence of physical and psychosocial factors on accuracy of memory for pain in chronic pain patients.Pain.1989;37:289-294.WSmithMSaferEffects of present pain level on recall of chronic pain and medication use.Pain.1993;55:355-361.SAPearceSIsherwoodDHroudaPHRichardsonAErskineJSkinnerMemory and pain: tests of mood congruity and state dependent learning in experimentally induced and clinical pain.Pain.1990;43:187-193.LJBaumannLDCameronRSZimmermanHLeventhalIllness representations and matching labels with symptoms.Health Psychol.1989;8:449-469.JASkeltonRTCroyleMental Representation in Health and Illness.New York, NY: Springer-Verlag Inc; 1991.RCKesslerHUWittchenJAbelsonSZhaoMethodological issues in assessing psychiatric disorders with self-reports.In: Stone AA, Turkkan JS, Bachrach C, Jobe JB, Kurtzman HS, Cain VS, eds. The Science of Self-Report: Implications for Research and Practice.Mahwah, NJ: Lawrence Erlbaum Associates; 2000:229-255.PDMarshallMO'ConnorJPHodgkinsonThe perceived relationship between neck symptoms and precedent injury.Injury.1995;26:17-19.WMittenbergDVDiGiulioSPerrinAEBassSymptoms following mild head injury: expectation as aetiology.J Neurol Neurosurg Psychiatry.1992;55:200-204.CMcFarlandMRossNDeCourvilleWomen's theories of menstruation and biases in recall of menstrual symptoms.J Pers Soc Psychol.1989;57:522-531.RTCroyleGNSandeDenial and confirmatory search: paradoxical consequences of medical diagnosis.J Appl Soc Psychol.1988;18:473-490.JBJobeRTourangeauAFSmithContributions of survey research to the understanding of memory.Appl Cognit Psychol.1993;7:567-584.NCSchaefferAsking questions about threatening topics: a selective overview.In: Stone AA, Turkkan JS, Bachrach CA, Jobe JB, Kurtzman HS, Cain VS, eds. The Science of Self-Report: Implications for Research and Practice.Mahwah, NJ: Lawrence Erlbaum Associates; 2000:105-121.EFLoftusWMarburgerSince the eruption of Mt St Helens, has anyone beaten you up? improving the accuracy of retrospective reports with landmark events.Mem Cognit.1983;11:114-120.RTourangeauRemembering what happened: memory errors and survey reports.In: Stone AA, Turkkan JS, Bachrach CA, Jobe JB, Kurtzman HS, Cain VS, eds. The Science of Self-Report: Implications for Research and Practice.Mahwah, NJ: Lawrence Erlbaum Associates; 2000:29-48.EFLoftusGAFavaRetrieving multiple autobiographical memories.Soc Cognit.1985;3:280-295.Accepted for publication September 13, 2001.Corresponding author and reprints: Arthur J. Barsky, MD, Department of Psychiatry, Brigham and Women's Hospital, 75 Francis St, Boston, MA 02115.
JAMA Internal Medicine – American Medical Association
Published: May 13, 2002